[mirror_ubuntu-artful-kernel.git] / fs / ceph / locks.c

#include <linux/ceph/ceph_debug.h>

#include <linux/file.h>
#include <linux/namei.h>
#include <linux/random.h>

#include "super.h"
#include "mds_client.h"
#include <linux/ceph/pagelist.h>

static u64 lock_secret;
static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
                                         struct ceph_mds_request *req);

static inline u64 secure_addr(void *addr)
{
	u64 v = lock_secret ^ (u64)(unsigned long)addr;
	/*
	 * Set the most significant bit, so that MDS knows the 'owner'
	 * is sufficient to identify the owner of lock. (old code uses
	 * both 'owner' and 'pid')
	 */
	v |= (1ULL << 63);
	return v;
}

void __init ceph_flock_init(void)
{
	get_random_bytes(&lock_secret, sizeof(lock_secret));
}

/**
 * Implement fcntl and flock locking functions.
 */
static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
			     int cmd, u8 wait, struct file_lock *fl)
{
	struct inode *inode = file_inode(file);
	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	struct ceph_mds_request *req;
	int err;
	u64 length = 0;
	u64 owner;

	if (operation != CEPH_MDS_OP_SETFILELOCK || cmd == CEPH_LOCK_UNLOCK)
		wait = 0;

	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	if (IS_ERR(req))
		return PTR_ERR(req);
	req->r_inode = inode;
	ihold(inode);
	req->r_num_caps = 1;

	/* mds requires start and length rather than start and end */
	if (LLONG_MAX == fl->fl_end)
		length = 0;
	else
		length = fl->fl_end - fl->fl_start + 1;

	owner = secure_addr(fl->fl_owner);

	dout("ceph_lock_message: rule: %d, op: %d, owner: %llx, pid: %llu, "
	     "start: %llu, length: %llu, wait: %d, type: %d", (int)lock_type,
	     (int)operation, owner, (u64)fl->fl_pid, fl->fl_start, length,
	     wait, fl->fl_type);

	req->r_args.filelock_change.rule = lock_type;
	req->r_args.filelock_change.type = cmd;
	req->r_args.filelock_change.owner = cpu_to_le64(owner);
	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
	req->r_args.filelock_change.length = cpu_to_le64(length);
	req->r_args.filelock_change.wait = wait;

	if (wait)
		req->r_wait_for_completion = ceph_lock_wait_for_completion;

	err = ceph_mdsc_do_request(mdsc, inode, req);

	if (operation == CEPH_MDS_OP_GETFILELOCK) {
		fl->fl_pid = le64_to_cpu(req->r_reply_info.filelock_reply->pid);
		if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
			fl->fl_type = F_RDLCK;
		else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
			fl->fl_type = F_WRLCK;
		else
			fl->fl_type = F_UNLCK;

		fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
		length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
						 le64_to_cpu(req->r_reply_info.filelock_reply->length);
		if (length >= 1)
			fl->fl_end = length -1;
		else
			fl->fl_end = 0;

	}
	ceph_mdsc_put_request(req);
	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
	     "length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
	     (int)operation, (u64)fl->fl_pid, fl->fl_start,
	     length, wait, fl->fl_type, err);
	return err;
}

static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
                                         struct ceph_mds_request *req)
{
	struct ceph_mds_request *intr_req;
	struct inode *inode = req->r_inode;
	int err, lock_type;

	BUG_ON(req->r_op != CEPH_MDS_OP_SETFILELOCK);
	if (req->r_args.filelock_change.rule == CEPH_LOCK_FCNTL)
		lock_type = CEPH_LOCK_FCNTL_INTR;
	else if (req->r_args.filelock_change.rule == CEPH_LOCK_FLOCK)
		lock_type = CEPH_LOCK_FLOCK_INTR;
	else
		BUG_ON(1);
	BUG_ON(req->r_args.filelock_change.type == CEPH_LOCK_UNLOCK);

	err = wait_for_completion_interruptible(&req->r_completion);
	if (!err)
		return 0;

	dout("ceph_lock_wait_for_completion: request %llu was interrupted\n",
	     req->r_tid);

	mutex_lock(&mdsc->mutex);
	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
		err = 0;
	} else {
		/*
		 * ensure we aren't running concurrently with
		 * ceph_fill_trace or ceph_readdir_prepopulate, which
		 * rely on locks (dir mutex) held by our caller.
		 */
		mutex_lock(&req->r_fill_mutex);
		req->r_err = err;
		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
		mutex_unlock(&req->r_fill_mutex);

		if (!req->r_session) {
			// haven't sent the request
			err = 0;
		}
	}
	mutex_unlock(&mdsc->mutex);
	if (!err)
		return 0;

	intr_req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETFILELOCK,
					    USE_AUTH_MDS);
	if (IS_ERR(intr_req))
		return PTR_ERR(intr_req);

	intr_req->r_inode = inode;
	ihold(inode);
	intr_req->r_num_caps = 1;

	intr_req->r_args.filelock_change = req->r_args.filelock_change;
	intr_req->r_args.filelock_change.rule = lock_type;
	intr_req->r_args.filelock_change.type = CEPH_LOCK_UNLOCK;

	err = ceph_mdsc_do_request(mdsc, inode, intr_req);
	ceph_mdsc_put_request(intr_req);

	if (err && err != -ERESTARTSYS)
		return err;

	wait_for_completion_killable(&req->r_safe_completion);
	return 0;
}

/**
 * Attempt to set an fcntl lock.
 * For now, this just goes away to the server. Later it may be more awesome.
 */
int ceph_lock(struct file *file, int cmd, struct file_lock *fl)
{
	u8 lock_cmd;
	int err;
	u8 wait = 0;
	u16 op = CEPH_MDS_OP_SETFILELOCK;

	if (!(fl->fl_flags & FL_POSIX))
		return -ENOLCK;
	/* No mandatory locks */
	if (__mandatory_lock(file->f_mapping->host) && fl->fl_type != F_UNLCK)
		return -ENOLCK;

	dout("ceph_lock, fl_owner: %p", fl->fl_owner);

	/* set wait bit as appropriate, then make command as Ceph expects it*/
	if (IS_GETLK(cmd))
		op = CEPH_MDS_OP_GETFILELOCK;
	else if (IS_SETLKW(cmd))
		wait = 1;

	if (F_RDLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_SHARED;
	else if (F_WRLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_EXCL;
	else
		lock_cmd = CEPH_LOCK_UNLOCK;

	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
	if (!err) {
		if (op != CEPH_MDS_OP_GETFILELOCK) {
			dout("mds locked, locking locally");
			err = posix_lock_file(file, fl, NULL);
			if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
				/* undo! This should only happen if
				 * the kernel detects local
				 * deadlock. */
				ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
						  CEPH_LOCK_UNLOCK, 0, fl);
				dout("got %d on posix_lock_file, undid lock",
				     err);
			}
		}
	}
	return err;
}

int ceph_flock(struct file *file, int cmd, struct file_lock *fl)
{
	u8 lock_cmd;
	int err;
	u8 wait = 0;

	if (!(fl->fl_flags & FL_FLOCK))
		return -ENOLCK;
	/* No mandatory locks */
	if (fl->fl_type & LOCK_MAND)
		return -EOPNOTSUPP;

	dout("ceph_flock, fl_file: %p", fl->fl_file);

	if (IS_SETLKW(cmd))
		wait = 1;

	if (F_RDLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_SHARED;
	else if (F_WRLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_EXCL;
	else
		lock_cmd = CEPH_LOCK_UNLOCK;

	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
				file, lock_cmd, wait, fl);
	if (!err) {
		err = locks_lock_file_wait(file, fl);
		if (err) {
			ceph_lock_message(CEPH_LOCK_FLOCK,
					  CEPH_MDS_OP_SETFILELOCK,
					  file, CEPH_LOCK_UNLOCK, 0, fl);
			dout("got %d on locks_lock_file_wait, undid lock", err);
		}
	}
	return err;
}

/*
 * Fills in the passed counter variables, so you can prepare pagelist metadata
 * before calling ceph_encode_locks.
 */
void ceph_count_locks(struct inode *inode, int *fcntl_count, int *flock_count)
{
	struct file_lock *lock;
	struct file_lock_context *ctx;

	*fcntl_count = 0;
	*flock_count = 0;

	ctx = inode->i_flctx;
	if (ctx) {
		spin_lock(&ctx->flc_lock);
		list_for_each_entry(lock, &ctx->flc_posix, fl_list)
			++(*fcntl_count);
		list_for_each_entry(lock, &ctx->flc_flock, fl_list)
			++(*flock_count);
		spin_unlock(&ctx->flc_lock);
	}
	dout("counted %d flock locks and %d fcntl locks",
	     *flock_count, *fcntl_count);
}

/**
 * Encode the flock and fcntl locks for the given inode into the ceph_filelock
 * array. Must be called with inode->i_lock already held.
 * If we encounter more of a specific lock type than expected, return -ENOSPC.
 */
int ceph_encode_locks_to_buffer(struct inode *inode,
				struct ceph_filelock *flocks,
				int num_fcntl_locks, int num_flock_locks)
{
	struct file_lock *lock;
	struct file_lock_context *ctx = inode->i_flctx;
	int err = 0;
	int seen_fcntl = 0;
	int seen_flock = 0;
	int l = 0;

	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	     num_fcntl_locks);

	if (!ctx)
		return 0;

	spin_lock(&ctx->flc_lock);
	list_for_each_entry(lock, &ctx->flc_posix, fl_list) {
		++seen_fcntl;
		if (seen_fcntl > num_fcntl_locks) {
			err = -ENOSPC;
			goto fail;
		}
		err = lock_to_ceph_filelock(lock, &flocks[l]);
		if (err)
			goto fail;
		++l;
	}
	list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
		++seen_flock;
		if (seen_flock > num_flock_locks) {
			err = -ENOSPC;
			goto fail;
		}
		err = lock_to_ceph_filelock(lock, &flocks[l]);
		if (err)
			goto fail;
		++l;
	}
fail:
	spin_unlock(&ctx->flc_lock);
	return err;
}

/**
 * Copy the encoded flock and fcntl locks into the pagelist.
 * Format is: #fcntl locks, sequential fcntl locks, #flock locks,
 * sequential flock locks.
 * Returns zero on success.
 */
int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
			   struct ceph_pagelist *pagelist,
			   int num_fcntl_locks, int num_flock_locks)
{
	int err = 0;
	__le32 nlocks;

	nlocks = cpu_to_le32(num_fcntl_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto out_fail;

	err = ceph_pagelist_append(pagelist, flocks,
				   num_fcntl_locks * sizeof(*flocks));
	if (err)
		goto out_fail;

	nlocks = cpu_to_le32(num_flock_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto out_fail;

	err = ceph_pagelist_append(pagelist,
				   &flocks[num_fcntl_locks],
				   num_flock_locks * sizeof(*flocks));
out_fail:
	return err;
}

/*
 * Given a pointer to a lock, convert it to a ceph filelock
 */
int lock_to_ceph_filelock(struct file_lock *lock,
			  struct ceph_filelock *cephlock)
{
	int err = 0;
	cephlock->start = cpu_to_le64(lock->fl_start);
	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	cephlock->client = cpu_to_le64(0);
	cephlock->pid = cpu_to_le64((u64)lock->fl_pid);
	cephlock->owner = cpu_to_le64(secure_addr(lock->fl_owner));

	switch (lock->fl_type) {
	case F_RDLCK:
		cephlock->type = CEPH_LOCK_SHARED;
		break;
	case F_WRLCK:
		cephlock->type = CEPH_LOCK_EXCL;
		break;
	case F_UNLCK:
		cephlock->type = CEPH_LOCK_UNLOCK;
		break;
	default:
		dout("Have unknown lock type %d", lock->fl_type);
		err = -EINVAL;
	}

	return err;
}
Commit	Line	Data
3d14c5d2	1	#include <linux/ceph/ceph_debug.h>
40819f6f GF	2
	3	#include <linux/file.h>
	4	#include <linux/namei.h>
eb13e832	5	#include <linux/random.h>
40819f6f GF	6
	7	#include "super.h"
	8	#include "mds_client.h"
3d14c5d2	9	#include <linux/ceph/pagelist.h>
40819f6f	10
eb13e832	11	static u64 lock_secret;
9280be24 YZ	12	static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
9280be24 YZ	13	struct ceph_mds_request *req);
eb13e832 YZ	14
	15	static inline u64 secure_addr(void *addr)
	16	{
	17	u64 v = lock_secret ^ (u64)(unsigned long)addr;
	18	/*
	19	* Set the most significant bit, so that MDS knows the 'owner'
	20	* is sufficient to identify the owner of lock. (old code uses
	21	* both 'owner' and 'pid')
	22	*/
	23	v \|= (1ULL << 63);
	24	return v;
	25	}
	26
	27	void __init ceph_flock_init(void)
	28	{
	29	get_random_bytes(&lock_secret, sizeof(lock_secret));
	30	}
	31
40819f6f GF	32	/**
	33	* Implement fcntl and flock locking functions.
	34	*/
	35	static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
637ae8d5	36	int cmd, u8 wait, struct file_lock *fl)
40819f6f	37	{
496ad9aa	38	struct inode *inode = file_inode(file);
eb13e832	39	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
40819f6f GF	40	struct ceph_mds_request *req;
40819f6f GF	41	int err;
637ae8d5	42	u64 length = 0;
eb13e832	43	u64 owner;
40819f6f	44
9280be24 YZ	45	if (operation != CEPH_MDS_OP_SETFILELOCK \|\| cmd == CEPH_LOCK_UNLOCK)
	46	wait = 0;
	47
40819f6f GF	48	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	49	if (IS_ERR(req))
	50	return PTR_ERR(req);
70b666c3 SW	51	req->r_inode = inode;
70b666c3 SW	52	ihold(inode);
3bd58143	53	req->r_num_caps = 1;
40819f6f	54
637ae8d5 HS	55	/* mds requires start and length rather than start and end */
	56	if (LLONG_MAX == fl->fl_end)
	57	length = 0;
	58	else
	59	length = fl->fl_end - fl->fl_start + 1;
	60
130d1f95	61	owner = secure_addr(fl->fl_owner);
eb13e832 YZ	62
	63	dout("ceph_lock_message: rule: %d, op: %d, owner: %llx, pid: %llu, "
	64	"start: %llu, length: %llu, wait: %d, type: %d", (int)lock_type,
	65	(int)operation, owner, (u64)fl->fl_pid, fl->fl_start, length,
	66	wait, fl->fl_type);
637ae8d5	67
40819f6f GF	68	req->r_args.filelock_change.rule = lock_type;
40819f6f GF	69	req->r_args.filelock_change.type = cmd;
eb13e832	70	req->r_args.filelock_change.owner = cpu_to_le64(owner);
637ae8d5	71	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
637ae8d5	72	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
40819f6f GF	73	req->r_args.filelock_change.length = cpu_to_le64(length);
	74	req->r_args.filelock_change.wait = wait;
	75
9280be24 YZ	76	if (wait)
	77	req->r_wait_for_completion = ceph_lock_wait_for_completion;
	78
40819f6f	79	err = ceph_mdsc_do_request(mdsc, inode, req);
a5b10629	80
eb13e832	81	if (operation == CEPH_MDS_OP_GETFILELOCK) {
a5b10629 HS	82	fl->fl_pid = le64_to_cpu(req->r_reply_info.filelock_reply->pid);
	83	if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
	84	fl->fl_type = F_RDLCK;
	85	else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
	86	fl->fl_type = F_WRLCK;
	87	else
	88	fl->fl_type = F_UNLCK;
	89
	90	fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
	91	length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
	92	le64_to_cpu(req->r_reply_info.filelock_reply->length);
	93	if (length >= 1)
	94	fl->fl_end = length -1;
	95	else
	96	fl->fl_end = 0;
	97
	98	}
40819f6f GF	99	ceph_mdsc_put_request(req);
40819f6f GF	100	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
0c1f91f2	101	"length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
637ae8d5 HS	102	(int)operation, (u64)fl->fl_pid, fl->fl_start,
637ae8d5 HS	103	length, wait, fl->fl_type, err);
40819f6f GF	104	return err;
	105	}
	106
9280be24 YZ	107	static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
	108	struct ceph_mds_request *req)
	109	{
	110	struct ceph_mds_request *intr_req;
	111	struct inode *inode = req->r_inode;
	112	int err, lock_type;
	113
	114	BUG_ON(req->r_op != CEPH_MDS_OP_SETFILELOCK);
	115	if (req->r_args.filelock_change.rule == CEPH_LOCK_FCNTL)
	116	lock_type = CEPH_LOCK_FCNTL_INTR;
	117	else if (req->r_args.filelock_change.rule == CEPH_LOCK_FLOCK)
	118	lock_type = CEPH_LOCK_FLOCK_INTR;
	119	else
	120	BUG_ON(1);
	121	BUG_ON(req->r_args.filelock_change.type == CEPH_LOCK_UNLOCK);
	122
	123	err = wait_for_completion_interruptible(&req->r_completion);
	124	if (!err)
	125	return 0;
	126
	127	dout("ceph_lock_wait_for_completion: request %llu was interrupted\n",
	128	req->r_tid);
	129
92e57e62 YZ	130	mutex_lock(&mdsc->mutex);
	131	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
	132	err = 0;
	133	} else {
	134	/*
	135	* ensure we aren't running concurrently with
	136	* ceph_fill_trace or ceph_readdir_prepopulate, which
	137	* rely on locks (dir mutex) held by our caller.
	138	*/
	139	mutex_lock(&req->r_fill_mutex);
	140	req->r_err = err;
	141	set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
	142	mutex_unlock(&req->r_fill_mutex);
	143
	144	if (!req->r_session) {
	145	// haven't sent the request
	146	err = 0;
	147	}
	148	}
	149	mutex_unlock(&mdsc->mutex);
	150	if (!err)
	151	return 0;
	152
9280be24 YZ	153	intr_req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETFILELOCK,
	154	USE_AUTH_MDS);
	155	if (IS_ERR(intr_req))
	156	return PTR_ERR(intr_req);
	157
	158	intr_req->r_inode = inode;
	159	ihold(inode);
	160	intr_req->r_num_caps = 1;
	161
	162	intr_req->r_args.filelock_change = req->r_args.filelock_change;
	163	intr_req->r_args.filelock_change.rule = lock_type;
	164	intr_req->r_args.filelock_change.type = CEPH_LOCK_UNLOCK;
	165
	166	err = ceph_mdsc_do_request(mdsc, inode, intr_req);
	167	ceph_mdsc_put_request(intr_req);
	168
	169	if (err && err != -ERESTARTSYS)
	170	return err;
	171
92e57e62	172	wait_for_completion_killable(&req->r_safe_completion);
9280be24 YZ	173	return 0;
	174	}
	175
40819f6f GF	176	/**
	177	* Attempt to set an fcntl lock.
	178	* For now, this just goes away to the server. Later it may be more awesome.
	179	*/
	180	int ceph_lock(struct file file, int cmd, struct file_lock fl)
	181	{
40819f6f GF	182	u8 lock_cmd;
	183	int err;
	184	u8 wait = 0;
	185	u16 op = CEPH_MDS_OP_SETFILELOCK;
	186
eb70c0ce YZ	187	if (!(fl->fl_flags & FL_POSIX))
	188	return -ENOLCK;
	189	/* No mandatory locks */
	190	if (__mandatory_lock(file->f_mapping->host) && fl->fl_type != F_UNLCK)
	191	return -ENOLCK;
	192
eb13e832	193	dout("ceph_lock, fl_owner: %p", fl->fl_owner);
40819f6f GF	194
40819f6f GF	195	/* set wait bit as appropriate, then make command as Ceph expects it*/
0e8e95d6	196	if (IS_GETLK(cmd))
40819f6f	197	op = CEPH_MDS_OP_GETFILELOCK;
0e8e95d6 YZ	198	else if (IS_SETLKW(cmd))
0e8e95d6 YZ	199	wait = 1;
40819f6f GF	200
	201	if (F_RDLCK == fl->fl_type)
	202	lock_cmd = CEPH_LOCK_SHARED;
	203	else if (F_WRLCK == fl->fl_type)
	204	lock_cmd = CEPH_LOCK_EXCL;
	205	else
	206	lock_cmd = CEPH_LOCK_UNLOCK;
	207
637ae8d5	208	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
40819f6f	209	if (!err) {
eb13e832	210	if (op != CEPH_MDS_OP_GETFILELOCK) {
a5b10629 HS	211	dout("mds locked, locking locally");
	212	err = posix_lock_file(file, fl, NULL);
	213	if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
0c1f91f2 SW	214	/* undo! This should only happen if
	215	* the kernel detects local
	216	* deadlock. */
a5b10629 HS	217	ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
a5b10629 HS	218	CEPH_LOCK_UNLOCK, 0, fl);
0c1f91f2 SW	219	dout("got %d on posix_lock_file, undid lock",
0c1f91f2 SW	220	err);
a5b10629	221	}
40819f6f	222	}
40819f6f GF	223	}
	224	return err;
	225	}
	226
	227	int ceph_flock(struct file file, int cmd, struct file_lock fl)
	228	{
40819f6f GF	229	u8 lock_cmd;
40819f6f GF	230	int err;
0e8e95d6	231	u8 wait = 0;
40819f6f	232
eb70c0ce YZ	233	if (!(fl->fl_flags & FL_FLOCK))
	234	return -ENOLCK;
	235	/* No mandatory locks */
db4a63aa YZ	236	if (fl->fl_type & LOCK_MAND)
db4a63aa YZ	237	return -EOPNOTSUPP;
eb70c0ce	238
eb13e832	239	dout("ceph_flock, fl_file: %p", fl->fl_file);
40819f6f	240
0e8e95d6 YZ	241	if (IS_SETLKW(cmd))
	242	wait = 1;
	243
	244	if (F_RDLCK == fl->fl_type)
40819f6f	245	lock_cmd = CEPH_LOCK_SHARED;
0e8e95d6	246	else if (F_WRLCK == fl->fl_type)
40819f6f GF	247	lock_cmd = CEPH_LOCK_EXCL;
	248	else
	249	lock_cmd = CEPH_LOCK_UNLOCK;
40819f6f GF	250
40819f6f GF	251	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
637ae8d5	252	file, lock_cmd, wait, fl);
40819f6f	253	if (!err) {
4f656367	254	err = locks_lock_file_wait(file, fl);
40819f6f GF	255	if (err) {
	256	ceph_lock_message(CEPH_LOCK_FLOCK,
	257	CEPH_MDS_OP_SETFILELOCK,
637ae8d5	258	file, CEPH_LOCK_UNLOCK, 0, fl);
4f656367	259	dout("got %d on locks_lock_file_wait, undid lock", err);
40819f6f	260	}
40819f6f GF	261	}
	262	return err;
	263	}
	264
5263e31e JL	265	/*
	266	* Fills in the passed counter variables, so you can prepare pagelist metadata
	267	* before calling ceph_encode_locks.
40819f6f GF	268	*/
	269	void ceph_count_locks(struct inode inode, int fcntl_count, int *flock_count)
	270	{
e084c1bd	271	struct file_lock *lock;
5263e31e	272	struct file_lock_context *ctx;
40819f6f GF	273
	274	*fcntl_count = 0;
	275	*flock_count = 0;
	276
5263e31e JL	277	ctx = inode->i_flctx;
5263e31e JL	278	if (ctx) {
e084c1bd JL	279	spin_lock(&ctx->flc_lock);
	280	list_for_each_entry(lock, &ctx->flc_posix, fl_list)
	281	++(*fcntl_count);
	282	list_for_each_entry(lock, &ctx->flc_flock, fl_list)
	283	++(*flock_count);
	284	spin_unlock(&ctx->flc_lock);
40819f6f GF	285	}
	286	dout("counted %d flock locks and %d fcntl locks",
	287	flock_count, fcntl_count);
	288	}
	289
	290	/**
39be95e9	291	* Encode the flock and fcntl locks for the given inode into the ceph_filelock
1c8c601a	292	* array. Must be called with inode->i_lock already held.
39be95e9	293	* If we encounter more of a specific lock type than expected, return -ENOSPC.
40819f6f	294	*/
39be95e9 JS	295	int ceph_encode_locks_to_buffer(struct inode *inode,
	296	struct ceph_filelock *flocks,
	297	int num_fcntl_locks, int num_flock_locks)
40819f6f GF	298	{
40819f6f GF	299	struct file_lock *lock;
bd61e0a9	300	struct file_lock_context *ctx = inode->i_flctx;
40819f6f	301	int err = 0;
fca4451a GF	302	int seen_fcntl = 0;
fca4451a GF	303	int seen_flock = 0;
39be95e9	304	int l = 0;
40819f6f GF	305
	306	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	307	num_fcntl_locks);
39be95e9	308
bd61e0a9 JL	309	if (!ctx)
	310	return 0;
	311
6109c850	312	spin_lock(&ctx->flc_lock);
f6762cb2	313	list_for_each_entry(lock, &ctx->flc_posix, fl_list) {
bd61e0a9 JL	314	++seen_fcntl;
	315	if (seen_fcntl > num_fcntl_locks) {
	316	err = -ENOSPC;
	317	goto fail;
40819f6f	318	}
bd61e0a9 JL	319	err = lock_to_ceph_filelock(lock, &flocks[l]);
	320	if (err)
	321	goto fail;
	322	++l;
40819f6f	323	}
bd61e0a9 JL	324	list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
	325	++seen_flock;
	326	if (seen_flock > num_flock_locks) {
	327	err = -ENOSPC;
	328	goto fail;
40819f6f	329	}
bd61e0a9 JL	330	err = lock_to_ceph_filelock(lock, &flocks[l]);
	331	if (err)
	332	goto fail;
	333	++l;
40819f6f GF	334	}
40819f6f GF	335	fail:
6109c850	336	spin_unlock(&ctx->flc_lock);
40819f6f GF	337	return err;
	338	}
	339
39be95e9 JS	340	/**
	341	* Copy the encoded flock and fcntl locks into the pagelist.
	342	* Format is: #fcntl locks, sequential fcntl locks, #flock locks,
	343	* sequential flock locks.
	344	* Returns zero on success.
	345	*/
	346	int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
	347	struct ceph_pagelist *pagelist,
	348	int num_fcntl_locks, int num_flock_locks)
	349	{
	350	int err = 0;
	351	__le32 nlocks;
	352
	353	nlocks = cpu_to_le32(num_fcntl_locks);
	354	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	355	if (err)
	356	goto out_fail;
	357
	358	err = ceph_pagelist_append(pagelist, flocks,
	359	num_fcntl_locks * sizeof(*flocks));
	360	if (err)
	361	goto out_fail;
	362
	363	nlocks = cpu_to_le32(num_flock_locks);
	364	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	365	if (err)
	366	goto out_fail;
	367
	368	err = ceph_pagelist_append(pagelist,
	369	&flocks[num_fcntl_locks],
	370	num_flock_locks * sizeof(*flocks));
	371	out_fail:
	372	return err;
	373	}
	374
40819f6f GF	375	/*
	376	* Given a pointer to a lock, convert it to a ceph filelock
	377	*/
	378	int lock_to_ceph_filelock(struct file_lock *lock,
	379	struct ceph_filelock *cephlock)
	380	{
	381	int err = 0;
40819f6f GF	382	cephlock->start = cpu_to_le64(lock->fl_start);
	383	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	384	cephlock->client = cpu_to_le64(0);
eb13e832	385	cephlock->pid = cpu_to_le64((u64)lock->fl_pid);
130d1f95	386	cephlock->owner = cpu_to_le64(secure_addr(lock->fl_owner));
40819f6f GF	387
	388	switch (lock->fl_type) {
	389	case F_RDLCK:
	390	cephlock->type = CEPH_LOCK_SHARED;
	391	break;
	392	case F_WRLCK:
	393	cephlock->type = CEPH_LOCK_EXCL;
	394	break;
	395	case F_UNLCK:
	396	cephlock->type = CEPH_LOCK_UNLOCK;
	397	break;
	398	default:
	399	dout("Have unknown lock type %d", lock->fl_type);
	400	err = -EINVAL;
	401	}
	402
	403	return err;
	404	}